Automating HPC Measurement with Hackystat

1
Automating HPC Measurement with Hackystat
Philip Johnson and Michael Paulding, University
of Hawaii, Honolulu, Hawaii.
Hackystat Developer Services Web Site

• What is Software Project Telemetry?
• Software Project Telemetry is a style of software
  metrics definition, collection, and analysis
  emphasizing
• Automated collection of metrics by attaching
  sensors to development environment tools
• Metrics are time-stamped events, providing data
  about either product or process at a particular
  moment
• Analyses emphasize trends in measures at various
  grain-sizes (days, weeks, months)
• Telemetry data provides in-process feedback to
  developers and managers, supporting improvement

What is Hackystat? The Hackystat research
project explores ways to provide software
development individuals and teams with automated
support for collecting and analyzing interesting
and useful measures of the process and products
of software development. Hackystat attempts to
decrease developer overhead for metrics
collection to zero by providing sensors for
development tools that unobtrusively collect and
send metrics data to a centralized web server.
Example Hackystat Telemetry Report

• Hackystat Feature Set
• Sensors Eclipse, Emacs, JBuilder, Vim, Jira,
  Visual Studio, Ant, JUnit, JBlanket, CCCC, BCML,
  Harvest, LOCC, Office, Unix CLI, CVS, CppUnit
• Sensor Data Types Activity, Coverage, UnitTest,
  FileMetric, Build, StateChange, Commit, CLI,
  Perf, Issue, Perf
• Analysis Types Personal Metrics, Project
  Metrics, Software Telemetry, Software Development
  Stream Analysis, Review, GQM
• Languages Java, C, C, Lisp

Example Hackystat Analyses Page

• High Performance Computing Projects with
  Hackystat
• University of Hawaii, Case study with Optimal
  Truss PBB
• University of Maryland, Student case studies of
  OpenMP/MPI
• Sun Microsystems, Professional developer case
  study
• Vanderbilt University, Eclipse IDE extensions
  for HPC

• For more information
• Beyond the Personal Software Process Metrics
  collection and analysis for the differently
  disciplined, P. Johnson et al, Proceedings of the
  2003 International Conference on Software
  Engineering.
• Improving Software Development Managaement
  through Software Project Telemetry, P. Johnson
  et. al, IEEE Software, August 2005
• http//www.hackystat.org/

Thanks to our sponsors